Convolutional Neural Networks

Project: Write an Algorithm for a Dog Identification App

---

In this notebook, some template code has already been provided for you, and you will need to implement additional functionality to successfully complete this project. You will not need to modify the included code beyond what is requested. Sections that begin with '(IMPLEMENTATION)' in the header indicate that the following block of code will require additional functionality which you must provide. Instructions will be provided for each section, and the specifics of the implementation are marked in the code block with a 'TODO' statement. Please be sure to read the instructions carefully!

Note: Once you have completed all of the code implementations, you need to finalize your work by exporting the Jupyter Notebook as an HTML document. Before exporting the notebook to html, all of the code cells need to have been run so that reviewers can see the final implementation and output. You can then export the notebook by using the menu above and navigating to File -> Download as -> HTML (.html). Include the finished document along with this notebook as your submission.

In addition to implementing code, there will be questions that you must answer which relate to the project and your implementation. Each section where you will answer a question is preceded by a 'Question X' header. Carefully read each question and provide thorough answers in the following text boxes that begin with 'Answer:'. Your project submission will be evaluated based on your answers to each of the questions and the implementation you provide.

Note: Code and Markdown cells can be executed using the Shift + Enter keyboard shortcut. Markdown cells can be edited by double-clicking the cell to enter edit mode.

The rubric contains optional "Stand Out Suggestions" for enhancing the project beyond the minimum requirements. If you decide to pursue the "Stand Out Suggestions", you should include the code in this Jupyter notebook.

---

Why We're Here

In this notebook, you will make the first steps towards developing an algorithm that could be used as part of a mobile or web app. At the end of this project, your code will accept any user-supplied image as input. If a dog is detected in the image, it will provide an estimate of the dog's breed. If a human is detected, it will provide an estimate of the dog breed that is most resembling. The image below displays potential sample output of your finished project (... but we expect that each student's algorithm will behave differently!).

In this real-world setting, you will need to piece together a series of models to perform different tasks; for instance, the algorithm that detects humans in an image will be different from the CNN that infers dog breed. There are many points of possible failure, and no perfect algorithm exists. Your imperfect solution will nonetheless create a fun user experience!

The Road Ahead

We break the notebook into separate steps. Feel free to use the links below to navigate the notebook.

• Step 0: Import Datasets
• Step 1: Detect Humans
• Step 2: Detect Dogs
• Step 3: Create a CNN to Classify Dog Breeds (from Scratch)
• Step 4: Create a CNN to Classify Dog Breeds (using Transfer Learning)
• Step 5: Write your Algorithm
• Step 6: Test Your Algorithm

---

Step 0: Import Datasets

Make sure that you've downloaded the required human and dog datasets:

• Download the dog dataset. Unzip the folder and place it in this project's home directory, at the location /dogImages.

• Download the human dataset. Unzip the folder and place it in the home directory, at location /lfw.

Note: If you are using a Windows machine, you are encouraged to use 7zip to extract the folder.

In the code cell below, we save the file paths for both the human (LFW) dataset and dog dataset in the numpy arrays human_files and dog_files.

import numpy as np
from glob import glob

# load filenames for human and dog images
human_files = np.array(glob("lfw/*/*/*"))
dog_files = np.array(glob("dogImages/*/*/*"))

# print number of images in each dataset
print('There are %d total human images.' % len(human_files))
print('There are %d total dog images.' % len(dog_files))

There are 18982 total human images.
There are 8351 total dog images.

Step 1: Detect Humans

In this section, we use OpenCV's implementation of Haar feature-based cascade classifiers to detect human faces in images.

OpenCV provides many pre-trained face detectors, stored as XML files on github. We have downloaded one of these detectors and stored it in the haarcascades directory. In the next code cell, we demonstrate how to use this detector to find human faces in a sample image.

#!pip install opencv-python
#!pip install cv2-tools

import cv2                
import matplotlib.pyplot as plt                        
%matplotlib inline                               

# extract pre-trained face detector
face_cascade = cv2.CascadeClassifier('haarcascades/haarcascade_frontalface_alt.xml')

# load color (BGR) image
img = cv2.imread(human_files[1])
# convert BGR image to grayscale
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

# find faces in image
faces = face_cascade.detectMultiScale(gray)

# print number of faces detected in the image
print('Number of faces detected:', len(faces))

# get bounding box for each detected face
for (x,y,w,h) in faces:
    # add bounding box to color image
    cv2.rectangle(img,(x,y),(x+w,y+h),(255,0,0),2)
    
# convert BGR image to RGB for plotting
cv_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)

# display the image, along with bounding box
plt.imshow(cv_rgb)
plt.show()

Number of faces detected: 1

Before using any of the face detectors, it is standard procedure to convert the images to grayscale. The detectMultiScale function executes the classifier stored in face_cascade and takes the grayscale image as a parameter.

In the above code, faces is a numpy array of detected faces, where each row corresponds to a detected face. Each detected face is a 1D array with four entries that specifies the bounding box of the detected face. The first two entries in the array (extracted in the above code as x and y) specify the horizontal and vertical positions of the top left corner of the bounding box. The last two entries in the array (extracted here as w and h) specify the width and height of the box.

Write a Human Face Detector

We can use this procedure to write a function that returns True if a human face is detected in an image and False otherwise. This function, aptly named face_detector, takes a string-valued file path to an image as input and appears in the code block below.

# returns "True" if face is detected in image stored at img_path
def face_detector(img_path):
    img = cv2.imread(img_path)
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    faces = face_cascade.detectMultiScale(gray)
    return len(faces) > 0

(IMPLEMENTATION) Assess the Human Face Detector

Question 1: Use the code cell below to test the performance of the face_detector function.

• What percentage of the first 100 images in human_files have a detected human face?
• What percentage of the first 100 images in dog_files have a detected human face?

Ideally, we would like 100% of human images with a detected face and 0% of dog images with a detected face. You will see that our algorithm falls short of this goal, but still gives acceptable performance. We extract the file paths for the first 100 images from each of the datasets and store them in the numpy arrays human_files_short and dog_files_short.

Answer: (You can print out your results and/or write your percentages in this cell)

from tqdm import tqdm

human_files_short = human_files[:100]
dog_files_short = dog_files[:100]

#-#-# Do NOT modify the code above this line. #-#-#

count = 0
for i in human_files_short:
    if face_detector(i):
        count += 1
print(count/len(human_files_short)*100)

count = 0
for i in dog_files_short:
    if face_detector(i):
        count += 1
print(count/len(dog_files_short)*100)

## TODO: Test the performance of the face_detector algorithm 
## on the images in human_files_short and dog_files_short.

95.0
18.0

We suggest the face detector from OpenCV as a potential way to detect human images in your algorithm, but you are free to explore other approaches, especially approaches that make use of deep learning :). Please use the code cell below to design and test your own face detection algorithm. If you decide to pursue this optional task, report performance on human_files_short and dog_files_short.

### (Optional) 
### TODO: Test performance of another face detection algorithm.
### Feel free to use as many code cells as needed.

---

Step 2: Detect Dogs

In this section, we use a pre-trained model to detect dogs in images.

Obtain Pre-trained VGG-16 Model

The code cell below downloads the VGG-16 model, along with weights that have been trained on ImageNet, a very large, very popular dataset used for image classification and other vision tasks. ImageNet contains over 10 million URLs, each linking to an image containing an object from one of 1000 categories.

import torch
import torchvision.models as models

# define VGG16 model
VGG16 = models.vgg16(pretrained=True)

# check if CUDA is available
use_cuda = torch.cuda.is_available()

# move model to GPU if CUDA is available
if use_cuda:
    VGG16 = VGG16.cuda()

Given an image, this pre-trained VGG-16 model returns a prediction (derived from the 1000 possible categories in ImageNet) for the object that is contained in the image.

(IMPLEMENTATION) Making Predictions with a Pre-trained Model

In the next code cell, you will write a function that accepts a path to an image (such as 'dogImages/train/001.Affenpinscher/Affenpinscher_00001.jpg') as input and returns the index corresponding to the ImageNet class that is predicted by the pre-trained VGG-16 model. The output should always be an integer between 0 and 999, inclusive.

Before writing the function, make sure that you take the time to learn how to appropriately pre-process tensors for pre-trained models in the PyTorch documentation.

VGG16

VGG(
  (features): Sequential(
    (0): Conv2d(3, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (1): ReLU(inplace=True)
    (2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (3): ReLU(inplace=True)
    (4): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
    (5): Conv2d(64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (6): ReLU(inplace=True)
    (7): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (8): ReLU(inplace=True)
    (9): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
    (10): Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (11): ReLU(inplace=True)
    (12): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (13): ReLU(inplace=True)
    (14): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (15): ReLU(inplace=True)
    (16): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
    (17): Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (18): ReLU(inplace=True)
    (19): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (20): ReLU(inplace=True)
    (21): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (22): ReLU(inplace=True)
    (23): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
    (24): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (25): ReLU(inplace=True)
    (26): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (27): ReLU(inplace=True)
    (28): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))
    (29): ReLU(inplace=True)
    (30): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=False)
  )
  (avgpool): AdaptiveAvgPool2d(output_size=(7, 7))
  (classifier): Sequential(
    (0): Linear(in_features=25088, out_features=4096, bias=True)
    (1): ReLU(inplace=True)
    (2): Dropout(p=0.5, inplace=False)
    (3): Linear(in_features=4096, out_features=4096, bias=True)
    (4): ReLU(inplace=True)
    (5): Dropout(p=0.5, inplace=False)
    (6): Linear(in_features=4096, out_features=1000, bias=True)
  )
)

from PIL import Image
import torchvision.transforms as transforms

# Set PIL to be tolerant of image files that are truncated.
from PIL import ImageFile
ImageFile.LOAD_TRUNCATED_IMAGES = True

def VGG16_predict(img_path):
    '''
    Use pre-trained VGG-16 model to obtain index corresponding to 
    predicted ImageNet class for image at specified path
    
    Args:
        img_path: path to an image
        
    Returns:
        Index corresponding to VGG-16 model's prediction
    '''
    
    ## TODO: Complete the function.
    ## Load and pre-process an image from the given img_path
    ## Return the *index* of the predicted class for that image
    #data_transform = transforms.Compose([transforms.RandomResizedCrop(256), transforms.ToTensor()])
    
    data_transform = transforms.Compose([transforms.Resize(224),
                                transforms.CenterCrop(224),
                                transforms.ToTensor(), 
                                transforms.Normalize(mean=[0.485, 0.456, 0.406],
                                                          std=[0.229, 0.224, 0.225])])
    
    
    image = Image.open(img_path)

    tensor = data_transform(image)
    
    tensor = tensor.unsqueeze(0)
    
    if use_cuda:
        tensor = tensor.cuda()
        
    
    result = VGG16(tensor)
    
    class_result = float(torch.argmax(result))
    
    return class_result # predicted class index

teste = VGG16_predict('dogImages/train/012.Australian_shepherd/Australian_shepherd_00832.jpg')
teste

231.0

(IMPLEMENTATION) Write a Dog Detector

While looking at the dictionary, you will notice that the categories corresponding to dogs appear in an uninterrupted sequence and correspond to dictionary keys 151-268, inclusive, to include all categories from 'Chihuahua' to 'Mexican hairless'. Thus, in order to check to see if an image is predicted to contain a dog by the pre-trained VGG-16 model, we need only check if the pre-trained model predicts an index between 151 and 268 (inclusive).

Use these ideas to complete the dog_detector function below, which returns True if a dog is detected in an image (and False if not).

### returns "True" if a dog is detected in the image stored at img_path
def dog_detector(img_path):
    ## TODO: Complete the function.
    if VGG16_predict(img_path) in range(151, 269):
        return True
    else:
        return False
     # true/false

teste = dog_detector('dogImages/train/001.Affenpinscher/Affenpinscher_00001.jpg')
teste

True

(IMPLEMENTATION) Assess the Dog Detector

Question 2: Use the code cell below to test the performance of your dog_detector function.

• What percentage of the images in human_files_short have a detected dog?
• What percentage of the images in dog_files_short have a detected dog?

Answer:

### TODO: Test the performance of the dog_detector function
### on the images in human_files_short and dog_files_short.
count_dog = 0
for i in human_files_short:
    if dog_detector(i) == True:
        count_dog += 1
print(count_dog/len(human_files_short))

count_dog = 0
for i in dog_files_short:
    if dog_detector(i) == True:
        count_dog += 1
print(count_dog/len(dog_files_short))

0.01
1.0

We suggest VGG-16 as a potential network to detect dog images in your algorithm, but you are free to explore other pre-trained networks (such as Inception-v3, ResNet-50, etc). Please use the code cell below to test other pre-trained PyTorch models. If you decide to pursue this optional task, report performance on human_files_short and dog_files_short.

### (Optional) 
### TODO: Report the performance of another pre-trained network.
### Feel free to use as many code cells as needed.

---

Step 3: Create a CNN to Classify Dog Breeds (from Scratch)

Now that we have functions for detecting humans and dogs in images, we need a way to predict breed from images. In this step, you will create a CNN that classifies dog breeds. You must create your CNN from scratch (so, you can't use transfer learning yet!), and you must attain a test accuracy of at least 10%. In Step 4 of this notebook, you will have the opportunity to use transfer learning to create a CNN that attains greatly improved accuracy.

We mention that the task of assigning breed to dogs from images is considered exceptionally challenging. To see why, consider that even a human would have trouble distinguishing between a Brittany and a Welsh Springer Spaniel.

Brittany	Welsh Springer Spaniel

It is not difficult to find other dog breed pairs with minimal inter-class variation (for instance, Curly-Coated Retrievers and American Water Spaniels).

Curly-Coated Retriever	American Water Spaniel

Likewise, recall that labradors come in yellow, chocolate, and black. Your vision-based algorithm will have to conquer this high intra-class variation to determine how to classify all of these different shades as the same breed.

Yellow Labrador	Chocolate Labrador	Black Labrador

We also mention that random chance presents an exceptionally low bar: setting aside the fact that the classes are slightly imabalanced, a random guess will provide a correct answer roughly 1 in 133 times, which corresponds to an accuracy of less than 1%.

Remember that the practice is far ahead of the theory in deep learning. Experiment with many different architectures, and trust your intuition. And, of course, have fun!

(IMPLEMENTATION) Specify Data Loaders for the Dog Dataset

Use the code cell below to write three separate data loaders for the training, validation, and test datasets of dog images (located at dogImages/train, dogImages/valid, and dogImages/test, respectively). You may find this documentation on custom datasets to be a useful resource. If you are interested in augmenting your training and/or validation data, check out the wide variety of transforms!

### TODO: Write data loaders for training, validation, and test sets
## Specify appropriate transforms, and batch_sizes

import os
from torchvision import datasets
from PIL import ImageFile

ImageFile.LOAD_TRUNCATED_IMAGES = True

transform_test = transforms.Compose([transforms.Resize(224),
                                     transforms.CenterCrop(224),
                                     transforms.RandomHorizontalFlip(),
                                     transforms.RandomRotation(20),
                                     transforms.ToTensor(), 
                                     transforms.Normalize(mean=[0.485, 0.456, 0.406],
                                                          std=[0.229, 0.224, 0.225])])

transform = transforms.Compose([transforms.Resize(224),
                                transforms.CenterCrop(224),
                                transforms.ToTensor(), 
                                transforms.Normalize(mean=[0.485, 0.456, 0.406],
                                                          std=[0.229, 0.224, 0.225])])

dataset = {}
loader_scratch = {}

dataset['train'] = datasets.ImageFolder(os.path.join('dogImages/train'), transform_test)
dataset['test'] = datasets.ImageFolder(os.path.join('dogImages/test'), transform)
dataset['valid'] = datasets.ImageFolder(os.path.join('dogImages/valid'), transform)

loader_scratch['train'] = torch.utils.data.DataLoader(dataset['train'], batch_size = 32, shuffle=True, num_workers=2)
loader_scratch['test'] = torch.utils.data.DataLoader(dataset['test'], batch_size = 16)
loader_scratch['valid'] = torch.utils.data.DataLoader(dataset['valid'], batch_size = 16)

print(len(dataset['train']))
print(len(dataset['test']))
print(len(dataset['valid']))
print(len(dataset['test'].classes))

6680
836
835
133

inputs, classes = next(iter(loader_scratch['train']))

inputs[0].shape

torch.Size([3, 224, 224])

classes

tensor([  6, 103,  74,  46, 108,  77,  99,  75,  16,  61,  23,  97,  31, 124,
         10, 121,  43,   9,  75, 107, 111, 102,  73,  37,  20, 112, 131, 101,
         88, 116,  44,  14])

from torchvision import utils

def visualize_sample_images(inp):
    inp = inp.numpy().transpose((1, 2, 0))
    inp = inp * np.array((0.229, 0.224, 0.225)) + np.array((0.485, 0.456, 0.406))
    inp = np.clip(inp, 0, 1)
    
    fig = plt.figure(figsize=(60, 25))
    plt.axis('off')
    plt.imshow(inp)
    plt.pause(0.001)
    
# Get a batch of training data.    
inputs, classes = next(iter(loader_scratch['train']))

# Convert the batch to a grid.
grid = utils.make_grid(inputs, nrow=5)

# Display!
visualize_sample_images(grid)

Question 3: Describe your chosen procedure for preprocessing the data.

• How does your code resize the images (by cropping, stretching, etc)? What size did you pick for the input tensor, and why?

  R: I decided to use resize, random rotation of 25 and random flip so it gives a good variability of the kink of images we can find outside the training data. Altought i saw some examples on internet using RandomResizedCrop i couldnt understand the reason why using. I wen for the 224 pixels size because its a good starting point so we dont lose too much information on the pictures we are training and we can adapt to outside images easier. I added Centercrop later because not all the tensor were the same size i discovered i need them all to be the same size

• Did you decide to augment the dataset? If so, how (through translations, flips, rotations, etc)? If not, why not?

  R: I decided to user only flip and a 25 degree rotation. I think that covers up most of images we are going to work with.

Answer:

torch.cuda.empty_cache() ### (IMPLEMENTATION) Model Architecture

Create a CNN to classify dog breed. Use the template in the code cell below.

import torch.nn as nn
import torch.nn.functional as F

# define the CNN architecture
class Net(nn.Module):
    ### TODO: choose an architecture, and complete the class
    def __init__(self):
        super().__init__()
        self.conv1 = nn.Conv2d(3, 64, 3, stride = 1, padding=1) 
        self.conv2 = nn.Conv2d(64, 128, 3, stride = 1, padding =1)
        self.conv3 = nn.Conv2d(128, 256, 3, stride = 1, padding =1)
        self.conv4 = nn.Conv2d(256, 512, 3, stride = 1, padding =1)
        self.conv5 = nn.Conv2d(512, 512, 3, stride = 1, padding =1)
                
        
        self.pool = nn.MaxPool2d(2, 2)
        
        self.fc1 = nn.Linear(25088, 512)
        self.fc2 = nn.Linear(512, 512) 
        self.fc3 = nn.Linear(512, 133)
        
        self.dropout = nn.Dropout(0.25)
        
        
    def forward(self, x):

        ## Define forward behavior
        x = self.pool(F.relu(self.conv1(x)))
        x = self.pool(F.relu(self.conv2(x)))
        x = self.pool(F.relu(self.conv3(x)))
        x = self.pool(F.relu(self.conv4(x)))
        x = self.pool(F.relu(self.conv5(x)))
        x = x.view(-1, 25088)
        x = F.relu(self.fc1(x))
        x = self.dropout(x)
        x = F.relu(self.fc2(x))
        x = self.dropout(x)
        x = self.fc3(x)
        return x
        
        
        
        return x

#-#-# You do NOT have to modify the code below this line. #-#-#

# instantiate the CNN
model_scratch = Net()

# move tensors to GPU if CUDA is available
if use_cuda:
    model_scratch.cuda()

# trying to debug why the dimensions were not working

conv1 = nn.Conv2d(3, 32, 2, stride = 2, padding=0) # 112 pixes after conv, and 56 after pool
#56 x 56 x 36
conv2 = nn.Conv2d(32, 64, 2, stride = 2, padding =0) # 28 pixels after, 14 after pool
#14 x 14 x 64

pool = nn.MaxPool2d(2, 2)

fc1 = nn.Linear(14 * 14 * 64, 6000)
fc2 = nn.Linear(6000, 500) # 133 output since it is 133 classes
fc3 = nn.Linear(500, 133) 




tensor_example = inputs[0]
print(tensor_example.size())

# apparently the conv2 expects the n_samples in the tensor dimension (n_samples, channels, height, width) # e.g., (1000, 1, 224, 224)
# adding unsqueeze to include 1 in the tensor. After further debbu

tensor_example = tensor_example.unsqueeze(0)
print(tensor_example.size())

tensor_example = conv1(tensor_example)
print(tensor_example.size())

tensor_example = pool(tensor_example)
print(tensor_example.size())

tensor_example = conv2(tensor_example)
print(tensor_example.size())

tensor_example = pool(tensor_example)
print(tensor_example.size())

tensor_example = tensor_example.view(-1, 64 * 14 * 14)
print(tensor_example.size())

tensor_example = fc1(tensor_example)
print(tensor_example.size())

tensor_example = fc2(tensor_example)
print(tensor_example.size())

tensor_example = fc3(tensor_example)
print(tensor_example.size())

tensor_example = tensor_example.squeeze()
print(tensor_example.size())

print(tensor_example)

torch.Size([3, 224, 224])
torch.Size([1, 3, 224, 224])
torch.Size([1, 32, 112, 112])
torch.Size([1, 32, 56, 56])
torch.Size([1, 64, 28, 28])
torch.Size([1, 64, 14, 14])
torch.Size([1, 12544])
torch.Size([1, 6000])
torch.Size([1, 500])
torch.Size([1, 133])
torch.Size([133])
tensor([-0.2461,  0.0105, -0.1061,  0.0987, -0.1113,  0.0884,  0.0187, -0.1838,
        -0.0789, -0.0333,  0.0574,  0.0399, -0.0104, -0.1320, -0.1030, -0.0041,
         0.0073,  0.0269, -0.2221, -0.1248, -0.0450,  0.1023, -0.0495, -0.0451,
        -0.1712,  0.1907, -0.0465, -0.1432, -0.1048,  0.1256, -0.0241, -0.0539,
        -0.0243,  0.0406,  0.1276,  0.0239, -0.0601, -0.1146, -0.0064, -0.2978,
         0.0078, -0.0487, -0.0483,  0.0543,  0.0815,  0.0349,  0.0690, -0.1955,
        -0.0337,  0.1523,  0.1824,  0.0533, -0.0190, -0.1032,  0.0576, -0.0185,
        -0.0743, -0.1310, -0.0599,  0.0141,  0.0977, -0.0884, -0.0634, -0.1948,
        -0.0754,  0.0786,  0.0241,  0.0746, -0.0654, -0.1168, -0.0102, -0.0600,
         0.0216,  0.0460,  0.0357, -0.0063, -0.0271, -0.1156, -0.0743,  0.0823,
         0.1462,  0.2088, -0.0401,  0.0439,  0.1173, -0.0048, -0.0155,  0.0883,
         0.0759,  0.0595, -0.0444,  0.1688,  0.0373, -0.1232, -0.0604, -0.2350,
        -0.0723, -0.0989, -0.2364,  0.1597, -0.0617,  0.0247, -0.0835,  0.0356,
        -0.0357,  0.0582, -0.0592,  0.1415,  0.0691, -0.0106,  0.0669, -0.0368,
        -0.1525,  0.1155, -0.0034, -0.0877, -0.0355, -0.2069, -0.1248,  0.0347,
         0.0136,  0.0372,  0.1138, -0.0431, -0.1340, -0.0472,  0.0704,  0.0700,
         0.1103,  0.0074, -0.1603,  0.0585, -0.1485],
       grad_fn=<SqueezeBackward0>)

Question 4: Outline the steps you took to get to your final CNN architecture and your reasoning at each step.

Answer:

I first started with the cnn model that is in the example of the pytorch documentation and gave it some tests. After running 2 epochs the loss was always 0.00. My first guess was that the feature size was not matching between the layers so i tryed adjusting the layers to the formula:

Feature size = ((Image size + 2 * Padding size − Kernel size) / Stride)+1

After much strugling using the formula i decided with the above model that fit into 224 pixels input

(IMPLEMENTATION) Specify Loss Function and Optimizer

Use the next code cell to specify a loss function and optimizer. Save the chosen loss function as criterion_scratch, and the optimizer as optimizer_scratch below.

import torch.optim as optim

### TODO: select loss function
criterion_scratch = nn.CrossEntropyLoss()

### TODO: select optimizer
optimizer_scratch = optim.SGD(model_scratch.parameters(), lr=0.01)

(IMPLEMENTATION) Train and Validate the Model

Train and validate your model in the code cell below. Save the final model parameters at filepath 'model_scratch.pt'.

# the following import is required for training to be robust to truncated images
from PIL import ImageFile
ImageFile.LOAD_TRUNCATED_IMAGES = True

def train(n_epochs, loaders, model, optimizer, criterion, use_cuda, save_path):
    """returns trained model"""
    # initialize tracker for minimum validation loss
    valid_loss_min = np.Inf 
    
    for epoch in range(1, n_epochs+1):
        # initialize variables to monitor training and validation loss
        train_loss = 0.0
        valid_loss = 0.0
        
        ###################
        # train the model #
        ###################
        model.train()
        for batch_idx, (data, target) in enumerate(loaders['train']):
            # move to GPU
            if use_cuda:
                data, target = data.cuda(), target.cuda()
                
                
            #clear gradients
            optimizer.zero_grad()
            
            #forward pass
            output = model(data)
            #calculate loss
            loss = criterion(output, target)
            #compute gradient
            loss.backward()
            #update parameters
            optimizer.step()
            
            #calculate train loss
            train_loss += loss.item()*data.size(0)
            
  
        ######################    
        # validate the model #
        ######################
        model.eval()
        for batch_idx, (data, target) in enumerate(loaders['valid']):
            # move to GPU
            if use_cuda:
                data, target = data.cuda(), target.cuda()
            ## update the average validation loss
            
            #forward pass
            output = model(data)
            #calculate loss
            loss = criterion(output, target)
            ## update the average validation loss
            valid_loss += loss.item()*data.size(0)
            
        train_loss = train_loss/len(loaders['train'].dataset)
        valid_loss = valid_loss/len(loaders['valid'].dataset)

            
        # print training/validation statistics 
        print('Epoch: {} \tTraining Loss: {:.6f} \tValidation Loss: {:.6f}'.format(
            epoch, 
            train_loss,
            valid_loss
            ))
        
        ## TODO: save the model if validation loss has decreased
        if valid_loss < valid_loss_min:
            print('Validation loss decreased ({:.6f} --> {:.6f}).  Saving model ...'.format(
            valid_loss_min,
            valid_loss))
            torch.save(model.state_dict(), save_path)
            valid_loss_min = valid_loss
            
    # return trained model
    return model


# train the model
model_scratch = train(100, loader_scratch, model_scratch, optimizer_scratch, 
                      criterion_scratch, use_cuda, 'model_scratch.pt')

#load the model that got the best validation accuracy
model_scratch.load_state_dict(torch.load('model_scratch.pt'))

Epoch: 1 	Training Loss: 4.889392 	Validation Loss: 4.888515
Validation loss decreased (inf --> 4.888515).  Saving model ...
Epoch: 2 	Training Loss: 4.887668 	Validation Loss: 4.887023
Validation loss decreased (4.888515 --> 4.887023).  Saving model ...
Epoch: 3 	Training Loss: 4.885967 	Validation Loss: 4.885342
Validation loss decreased (4.887023 --> 4.885342).  Saving model ...
Epoch: 4 	Training Loss: 4.883670 	Validation Loss: 4.883016
Validation loss decreased (4.885342 --> 4.883016).  Saving model ...
Epoch: 5 	Training Loss: 4.880407 	Validation Loss: 4.878689
Validation loss decreased (4.883016 --> 4.878689).  Saving model ...
Epoch: 6 	Training Loss: 4.874915 	Validation Loss: 4.871521
Validation loss decreased (4.878689 --> 4.871521).  Saving model ...
Epoch: 7 	Training Loss: 4.867584 	Validation Loss: 4.865218
Validation loss decreased (4.871521 --> 4.865218).  Saving model ...
Epoch: 8 	Training Loss: 4.865508 	Validation Loss: 4.861340
Validation loss decreased (4.865218 --> 4.861340).  Saving model ...
Epoch: 9 	Training Loss: 4.860388 	Validation Loss: 4.852155
Validation loss decreased (4.861340 --> 4.852155).  Saving model ...
Epoch: 10 	Training Loss: 4.841278 	Validation Loss: 4.816055
Validation loss decreased (4.852155 --> 4.816055).  Saving model ...
Epoch: 11 	Training Loss: 4.794293 	Validation Loss: 4.749731
Validation loss decreased (4.816055 --> 4.749731).  Saving model ...
Epoch: 12 	Training Loss: 4.730763 	Validation Loss: 4.699506
Validation loss decreased (4.749731 --> 4.699506).  Saving model ...
Epoch: 13 	Training Loss: 4.664679 	Validation Loss: 4.637422
Validation loss decreased (4.699506 --> 4.637422).  Saving model ...
Epoch: 14 	Training Loss: 4.612943 	Validation Loss: 4.632593
Validation loss decreased (4.637422 --> 4.632593).  Saving model ...
Epoch: 15 	Training Loss: 4.580469 	Validation Loss: 4.548622
Validation loss decreased (4.632593 --> 4.548622).  Saving model ...
Epoch: 16 	Training Loss: 4.546736 	Validation Loss: 4.539577
Validation loss decreased (4.548622 --> 4.539577).  Saving model ...
Epoch: 17 	Training Loss: 4.512474 	Validation Loss: 4.550806
Epoch: 18 	Training Loss: 4.475593 	Validation Loss: 4.507340
Validation loss decreased (4.539577 --> 4.507340).  Saving model ...
Epoch: 19 	Training Loss: 4.451274 	Validation Loss: 4.494305
Validation loss decreased (4.507340 --> 4.494305).  Saving model ...
Epoch: 20 	Training Loss: 4.412459 	Validation Loss: 4.469170
Validation loss decreased (4.494305 --> 4.469170).  Saving model ...
Epoch: 21 	Training Loss: 4.385820 	Validation Loss: 4.463912
Validation loss decreased (4.469170 --> 4.463912).  Saving model ...
Epoch: 22 	Training Loss: 4.356483 	Validation Loss: 4.462934
Validation loss decreased (4.463912 --> 4.462934).  Saving model ...
Epoch: 23 	Training Loss: 4.318851 	Validation Loss: 4.425667
Validation loss decreased (4.462934 --> 4.425667).  Saving model ...
Epoch: 24 	Training Loss: 4.287295 	Validation Loss: 4.392981
Validation loss decreased (4.425667 --> 4.392981).  Saving model ...
Epoch: 25 	Training Loss: 4.258789 	Validation Loss: 4.360056
Validation loss decreased (4.392981 --> 4.360056).  Saving model ...
Epoch: 26 	Training Loss: 4.212999 	Validation Loss: 4.377210
Epoch: 27 	Training Loss: 4.184479 	Validation Loss: 4.375341
Epoch: 28 	Training Loss: 4.156984 	Validation Loss: 4.301292
Validation loss decreased (4.360056 --> 4.301292).  Saving model ...
Epoch: 29 	Training Loss: 4.114704 	Validation Loss: 4.239886
Validation loss decreased (4.301292 --> 4.239886).  Saving model ...
Epoch: 30 	Training Loss: 4.089867 	Validation Loss: 4.243351
Epoch: 31 	Training Loss: 4.047095 	Validation Loss: 4.266069
Epoch: 32 	Training Loss: 4.019708 	Validation Loss: 4.190802
Validation loss decreased (4.239886 --> 4.190802).  Saving model ...
Epoch: 33 	Training Loss: 3.992199 	Validation Loss: 4.241649
Epoch: 34 	Training Loss: 3.956684 	Validation Loss: 4.218399
Epoch: 35 	Training Loss: 3.914070 	Validation Loss: 4.212701
Epoch: 36 	Training Loss: 3.874903 	Validation Loss: 4.237349
Epoch: 37 	Training Loss: 3.839453 	Validation Loss: 4.090529
Validation loss decreased (4.190802 --> 4.090529).  Saving model ...
Epoch: 38 	Training Loss: 3.807030 	Validation Loss: 4.062075
Validation loss decreased (4.090529 --> 4.062075).  Saving model ...
Epoch: 39 	Training Loss: 3.775096 	Validation Loss: 4.068092
Epoch: 40 	Training Loss: 3.721448 	Validation Loss: 4.032248
Validation loss decreased (4.062075 --> 4.032248).  Saving model ...
Epoch: 41 	Training Loss: 3.672801 	Validation Loss: 4.063045
Epoch: 42 	Training Loss: 3.620872 	Validation Loss: 4.099548
Epoch: 43 	Training Loss: 3.567962 	Validation Loss: 4.006018
Validation loss decreased (4.032248 --> 4.006018).  Saving model ...
Epoch: 44 	Training Loss: 3.503641 	Validation Loss: 3.918586
Validation loss decreased (4.006018 --> 3.918586).  Saving model ...
Epoch: 45 	Training Loss: 3.472104 	Validation Loss: 3.974407
Epoch: 46 	Training Loss: 3.416370 	Validation Loss: 3.937512
Epoch: 47 	Training Loss: 3.332091 	Validation Loss: 3.916004
Validation loss decreased (3.918586 --> 3.916004).  Saving model ...
Epoch: 48 	Training Loss: 3.291141 	Validation Loss: 3.902017
Validation loss decreased (3.916004 --> 3.902017).  Saving model ...
Epoch: 49 	Training Loss: 3.205323 	Validation Loss: 3.861680
Validation loss decreased (3.902017 --> 3.861680).  Saving model ...
Epoch: 50 	Training Loss: 3.140106 	Validation Loss: 3.868181
Epoch: 51 	Training Loss: 3.068250 	Validation Loss: 3.861856
Epoch: 52 	Training Loss: 3.013263 	Validation Loss: 3.891488
Epoch: 53 	Training Loss: 2.935923 	Validation Loss: 3.833490
Validation loss decreased (3.861680 --> 3.833490).  Saving model ...
Epoch: 54 	Training Loss: 2.861135 	Validation Loss: 3.921787
Epoch: 55 	Training Loss: 2.773815 	Validation Loss: 3.848594
Epoch: 56 	Training Loss: 2.704450 	Validation Loss: 3.905088
Epoch: 57 	Training Loss: 2.636155 	Validation Loss: 3.950334
Epoch: 58 	Training Loss: 2.556215 	Validation Loss: 3.819920
Validation loss decreased (3.833490 --> 3.819920).  Saving model ...
Epoch: 59 	Training Loss: 2.472448 	Validation Loss: 3.968770
Epoch: 60 	Training Loss: 2.373638 	Validation Loss: 3.982664
Epoch: 61 	Training Loss: 2.270091 	Validation Loss: 3.957459
Epoch: 62 	Training Loss: 2.194291 	Validation Loss: 4.004487
Epoch: 63 	Training Loss: 2.128164 	Validation Loss: 3.972860
Epoch: 64 	Training Loss: 2.034451 	Validation Loss: 4.347200
Epoch: 65 	Training Loss: 1.930092 	Validation Loss: 4.110896
Epoch: 66 	Training Loss: 1.848131 	Validation Loss: 4.073692
Epoch: 67 	Training Loss: 1.772752 	Validation Loss: 4.272849
Epoch: 68 	Training Loss: 1.657077 	Validation Loss: 4.436198
Epoch: 69 	Training Loss: 1.639359 	Validation Loss: 4.272867
Epoch: 70 	Training Loss: 1.529748 	Validation Loss: 4.195291
Epoch: 71 	Training Loss: 1.459628 	Validation Loss: 4.642825
Epoch: 72 	Training Loss: 1.398450 	Validation Loss: 4.470085
Epoch: 73 	Training Loss: 1.307647 	Validation Loss: 4.226277
Epoch: 74 	Training Loss: 1.222971 	Validation Loss: 4.572246
Epoch: 75 	Training Loss: 1.197259 	Validation Loss: 4.646921
Epoch: 76 	Training Loss: 1.146386 	Validation Loss: 4.582765
Epoch: 77 	Training Loss: 1.030041 	Validation Loss: 4.751989
Epoch: 78 	Training Loss: 1.019715 	Validation Loss: 4.816722
Epoch: 79 	Training Loss: 0.936239 	Validation Loss: 4.613564
Epoch: 80 	Training Loss: 0.890672 	Validation Loss: 4.938927
Epoch: 81 	Training Loss: 0.915127 	Validation Loss: 4.892101
Epoch: 82 	Training Loss: 0.847747 	Validation Loss: 4.957171
Epoch: 83 	Training Loss: 0.796957 	Validation Loss: 5.154359
Epoch: 84 	Training Loss: 0.765211 	Validation Loss: 5.068325
Epoch: 85 	Training Loss: 0.731599 	Validation Loss: 5.171645
Epoch: 86 	Training Loss: 0.696405 	Validation Loss: 5.118921
Epoch: 87 	Training Loss: 0.665685 	Validation Loss: 5.239987
Epoch: 88 	Training Loss: 0.606736 	Validation Loss: 5.043240
Epoch: 89 	Training Loss: 0.637375 	Validation Loss: 5.094911
Epoch: 90 	Training Loss: 0.588122 	Validation Loss: 5.530138
Epoch: 91 	Training Loss: 0.581235 	Validation Loss: 5.304848
Epoch: 92 	Training Loss: 0.522850 	Validation Loss: 5.498141
Epoch: 93 	Training Loss: 0.531247 	Validation Loss: 5.303794
Epoch: 94 	Training Loss: 0.487821 	Validation Loss: 5.548544
Epoch: 95 	Training Loss: 0.488928 	Validation Loss: 5.361702
Epoch: 96 	Training Loss: 0.452610 	Validation Loss: 5.372768
Epoch: 97 	Training Loss: 0.416665 	Validation Loss: 5.673854
Epoch: 98 	Training Loss: 0.412517 	Validation Loss: 5.957154
Epoch: 99 	Training Loss: 0.416761 	Validation Loss: 5.583009
Epoch: 100 	Training Loss: 0.424272 	Validation Loss: 5.792906

<All keys matched successfully>

(IMPLEMENTATION) Test the Model

Try out your model on the test dataset of dog images. Use the code cell below to calculate and print the test loss and accuracy. Ensure that your test accuracy is greater than 10%.

def test(loaders, model, criterion, use_cuda):

    # monitor test loss and accuracy
    test_loss = 0.
    correct = 0.
    total = 0.

    model.eval()
    for batch_idx, (data, target) in enumerate(loaders['test']):
        # move to GPU
        if use_cuda:
            data, target = data.cuda(), target.cuda()
        # forward pass: compute predicted outputs by passing inputs to the model
        output = model(data)
        # calculate the loss
        loss = criterion(output, target)
        # update average test loss 
        test_loss = test_loss + ((1 / (batch_idx + 1)) * (loss.data - test_loss))
        # convert output probabilities to predicted class
        pred = output.data.max(1, keepdim=True)[1]
        # compare predictions to true label
        correct += np.sum(np.squeeze(pred.eq(target.data.view_as(pred))).cpu().numpy())
        total += data.size(0)
            
    print('Test Loss: {:.6f}\n'.format(test_loss))

    print('\nTest Accuracy: %2d%% (%2d/%2d)' % (
        100. * correct / total, correct, total))

# call test function    
test(loader_scratch, model_scratch, criterion_scratch, use_cuda)

Test Loss: 3.935294


Test Accuracy: 15% (129/836)

---

Step 4: Create a CNN to Classify Dog Breeds (using Transfer Learning)

You will now use transfer learning to create a CNN that can identify dog breed from images. Your CNN must attain at least 60% accuracy on the test set.

(IMPLEMENTATION) Specify Data Loaders for the Dog Dataset

Use the code cell below to write three separate data loaders for the training, validation, and test datasets of dog images (located at dogImages/train, dogImages/valid, and dogImages/test, respectively).

If you like, you are welcome to use the same data loaders from the previous step, when you created a CNN from scratch.

## TODO: Specify data loaders
import os
import torch
from torchvision import datasets
import torchvision.transforms as transforms
from PIL import ImageFile


ImageFile.LOAD_TRUNCATED_IMAGES = True

transform_test = transforms.Compose([transforms.Resize(224),
                                     transforms.CenterCrop(224),
                                     transforms.RandomHorizontalFlip(),
                                     transforms.RandomRotation(20),
                                     transforms.ToTensor(), 
                                     transforms.Normalize(mean=[0.485, 0.456, 0.406],
                                                          std=[0.229, 0.224, 0.225])])

transform = transforms.Compose([transforms.Resize(224),
                                transforms.CenterCrop(224),
                                transforms.ToTensor(), 
                                transforms.Normalize(mean=[0.485, 0.456, 0.406],
                                                          std=[0.229, 0.224, 0.225])])

dataset = {}
loaders_transfer = {}

dataset['train'] = datasets.ImageFolder(os.path.join('dogImages/train'), transform_test)
dataset['test'] = datasets.ImageFolder(os.path.join('dogImages/test'), transform)
dataset['valid'] = datasets.ImageFolder(os.path.join('dogImages/valid'), transform)

loaders_transfer['train'] = torch.utils.data.DataLoader(dataset['train'], batch_size = 15, shuffle=True, num_workers=0)
loaders_transfer['test'] = torch.utils.data.DataLoader(dataset['test'], batch_size = 10)
loaders_transfer['valid'] = torch.utils.data.DataLoader(dataset['valid'], batch_size = 10)

(IMPLEMENTATION) Model Architecture

Use transfer learning to create a CNN to classify dog breed. Use the code cell below, and save your initialized model as the variable model_transfer.

import torchvision.models as models
import torch.nn as nn

## TODO: Specify model architecture 

model_transfer = models.vgg16_bn(pretrained=True)

for param in model_transfer.parameters():
    param.requires_grad = False

model_transfer.classifier[6] = nn.Linear(4096, 133, bias=True)

for param in model_transfer.classifier[6].parameters():
    param.requires_grad = True

if use_cuda:
    model_transfer = model_transfer.cuda()

Question 5: Outline the steps you took to get to your final CNN architecture and your reasoning at each step. Describe why you think the architecture is suitable for the current problem.

Answer: I decided for the VGG16 because its a good architeture for image that has good parameters suitable for imagenet that alredy contain picture of dogs.

I froze all parameters but decided to train the ones in the classifier for myself to see the results

(IMPLEMENTATION) Specify Loss Function and Optimizer

Use the next code cell to specify a loss function and optimizer. Save the chosen loss function as criterion_transfer, and the optimizer as optimizer_transfer below.

import torch.optim as optim

criterion_transfer = nn.CrossEntropyLoss()
optimizer_transfer = optim.SGD(model_transfer.parameters(), lr=0.01)

(IMPLEMENTATION) Train and Validate the Model

Train and validate your model in the code cell below. Save the final model parameters at filepath 'model_transfer.pt'.

# train the model
model_transfer =  train(100, loaders_transfer, model_transfer, optimizer_transfer, criterion_transfer, use_cuda, 'model_transfer.pt')

# load the model that got the best validation accuracy (uncomment the line below)
model_transfer.load_state_dict(torch.load('model_transfer.pt'))

Epoch: 1 	Training Loss: 2.612879 	Validation Loss: 1.181017
Validation loss decreased (inf --> 1.181017).  Saving model ...
Epoch: 2 	Training Loss: 1.233017 	Validation Loss: 0.800645
Validation loss decreased (1.181017 --> 0.800645).  Saving model ...
Epoch: 3 	Training Loss: 0.936194 	Validation Loss: 0.674867
Validation loss decreased (0.800645 --> 0.674867).  Saving model ...
Epoch: 4 	Training Loss: 0.824795 	Validation Loss: 0.594347
Validation loss decreased (0.674867 --> 0.594347).  Saving model ...
Epoch: 5 	Training Loss: 0.727943 	Validation Loss: 0.564899
Validation loss decreased (0.594347 --> 0.564899).  Saving model ...
Epoch: 6 	Training Loss: 0.685034 	Validation Loss: 0.534739
Validation loss decreased (0.564899 --> 0.534739).  Saving model ...
Epoch: 7 	Training Loss: 0.656422 	Validation Loss: 0.541734
Epoch: 8 	Training Loss: 0.614956 	Validation Loss: 0.519602
Validation loss decreased (0.534739 --> 0.519602).  Saving model ...
Epoch: 9 	Training Loss: 0.582113 	Validation Loss: 0.507413
Validation loss decreased (0.519602 --> 0.507413).  Saving model ...
Epoch: 10 	Training Loss: 0.577854 	Validation Loss: 0.482938
Validation loss decreased (0.507413 --> 0.482938).  Saving model ...
Epoch: 11 	Training Loss: 0.562529 	Validation Loss: 0.467366
Validation loss decreased (0.482938 --> 0.467366).  Saving model ...
Epoch: 12 	Training Loss: 0.547566 	Validation Loss: 0.462992
Validation loss decreased (0.467366 --> 0.462992).  Saving model ...
Epoch: 13 	Training Loss: 0.519890 	Validation Loss: 0.455672
Validation loss decreased (0.462992 --> 0.455672).  Saving model ...
Epoch: 14 	Training Loss: 0.498376 	Validation Loss: 0.490099
Epoch: 15 	Training Loss: 0.500148 	Validation Loss: 0.454600
Validation loss decreased (0.455672 --> 0.454600).  Saving model ...
Epoch: 16 	Training Loss: 0.507216 	Validation Loss: 0.457928
Epoch: 17 	Training Loss: 0.495924 	Validation Loss: 0.457241
Epoch: 18 	Training Loss: 0.469488 	Validation Loss: 0.466734
Epoch: 19 	Training Loss: 0.474311 	Validation Loss: 0.465533
Epoch: 20 	Training Loss: 0.467176 	Validation Loss: 0.443478
Validation loss decreased (0.454600 --> 0.443478).  Saving model ...
Epoch: 21 	Training Loss: 0.446043 	Validation Loss: 0.430522
Validation loss decreased (0.443478 --> 0.430522).  Saving model ...
Epoch: 22 	Training Loss: 0.448426 	Validation Loss: 0.444070
Epoch: 23 	Training Loss: 0.434091 	Validation Loss: 0.418937
Validation loss decreased (0.430522 --> 0.418937).  Saving model ...
Epoch: 24 	Training Loss: 0.444278 	Validation Loss: 0.442802
Epoch: 25 	Training Loss: 0.412774 	Validation Loss: 0.439757
Epoch: 26 	Training Loss: 0.440968 	Validation Loss: 0.437896
Epoch: 27 	Training Loss: 0.421756 	Validation Loss: 0.425798
Epoch: 28 	Training Loss: 0.415110 	Validation Loss: 0.425328
Epoch: 29 	Training Loss: 0.413895 	Validation Loss: 0.430368
Epoch: 30 	Training Loss: 0.400309 	Validation Loss: 0.415237
Validation loss decreased (0.418937 --> 0.415237).  Saving model ...
Epoch: 31 	Training Loss: 0.406444 	Validation Loss: 0.438446
Epoch: 32 	Training Loss: 0.397863 	Validation Loss: 0.421616
Epoch: 33 	Training Loss: 0.406506 	Validation Loss: 0.421152
Epoch: 34 	Training Loss: 0.391738 	Validation Loss: 0.424920
Epoch: 35 	Training Loss: 0.367713 	Validation Loss: 0.415680
Epoch: 36 	Training Loss: 0.385051 	Validation Loss: 0.417934
Epoch: 37 	Training Loss: 0.375831 	Validation Loss: 0.417978
Epoch: 38 	Training Loss: 0.375965 	Validation Loss: 0.434339
Epoch: 39 	Training Loss: 0.368646 	Validation Loss: 0.400417
Validation loss decreased (0.415237 --> 0.400417).  Saving model ...
Epoch: 40 	Training Loss: 0.376518 	Validation Loss: 0.405187
Epoch: 41 	Training Loss: 0.371183 	Validation Loss: 0.425622
Epoch: 42 	Training Loss: 0.348367 	Validation Loss: 0.426709
Epoch: 43 	Training Loss: 0.367837 	Validation Loss: 0.437932
Epoch: 44 	Training Loss: 0.349732 	Validation Loss: 0.403636
Epoch: 45 	Training Loss: 0.351014 	Validation Loss: 0.428028
Epoch: 46 	Training Loss: 0.352534 	Validation Loss: 0.428419
Epoch: 47 	Training Loss: 0.361653 	Validation Loss: 0.408701
Epoch: 48 	Training Loss: 0.350573 	Validation Loss: 0.406849
Epoch: 49 	Training Loss: 0.347704 	Validation Loss: 0.412908
Epoch: 50 	Training Loss: 0.339198 	Validation Loss: 0.411927
Epoch: 51 	Training Loss: 0.336375 	Validation Loss: 0.417306
Epoch: 52 	Training Loss: 0.338202 	Validation Loss: 0.421846
Epoch: 53 	Training Loss: 0.351281 	Validation Loss: 0.415302
Epoch: 54 	Training Loss: 0.339228 	Validation Loss: 0.412877
Epoch: 55 	Training Loss: 0.334407 	Validation Loss: 0.434391
Epoch: 56 	Training Loss: 0.330468 	Validation Loss: 0.409621
Epoch: 57 	Training Loss: 0.323064 	Validation Loss: 0.410408
Epoch: 58 	Training Loss: 0.329464 	Validation Loss: 0.425064
Epoch: 59 	Training Loss: 0.336731 	Validation Loss: 0.416017
Epoch: 60 	Training Loss: 0.327516 	Validation Loss: 0.430951
Epoch: 61 	Training Loss: 0.313994 	Validation Loss: 0.410825
Epoch: 62 	Training Loss: 0.316981 	Validation Loss: 0.406483
Epoch: 63 	Training Loss: 0.328437 	Validation Loss: 0.418080
Epoch: 64 	Training Loss: 0.319917 	Validation Loss: 0.386174
Validation loss decreased (0.400417 --> 0.386174).  Saving model ...
Epoch: 65 	Training Loss: 0.317090 	Validation Loss: 0.394199
Epoch: 66 	Training Loss: 0.311354 	Validation Loss: 0.413747
Epoch: 67 	Training Loss: 0.318958 	Validation Loss: 0.411925
Epoch: 68 	Training Loss: 0.316144 	Validation Loss: 0.423007
Epoch: 69 	Training Loss: 0.310914 	Validation Loss: 0.416942
Epoch: 70 	Training Loss: 0.316879 	Validation Loss: 0.424328
Epoch: 71 	Training Loss: 0.320683 	Validation Loss: 0.413274
Epoch: 72 	Training Loss: 0.302789 	Validation Loss: 0.413938
Epoch: 73 	Training Loss: 0.311104 	Validation Loss: 0.426094
Epoch: 74 	Training Loss: 0.304994 	Validation Loss: 0.393382
Epoch: 75 	Training Loss: 0.307655 	Validation Loss: 0.429122
Epoch: 76 	Training Loss: 0.306521 	Validation Loss: 0.421371
Epoch: 77 	Training Loss: 0.296730 	Validation Loss: 0.398145
Epoch: 78 	Training Loss: 0.287782 	Validation Loss: 0.415668
Epoch: 79 	Training Loss: 0.287106 	Validation Loss: 0.402193
Epoch: 80 	Training Loss: 0.313072 	Validation Loss: 0.396561
Epoch: 81 	Training Loss: 0.292493 	Validation Loss: 0.387197
Epoch: 82 	Training Loss: 0.293730 	Validation Loss: 0.431003
Epoch: 83 	Training Loss: 0.286474 	Validation Loss: 0.400926
Epoch: 84 	Training Loss: 0.301340 	Validation Loss: 0.392804
Epoch: 85 	Training Loss: 0.288828 	Validation Loss: 0.412119
Epoch: 86 	Training Loss: 0.295847 	Validation Loss: 0.399132
Epoch: 87 	Training Loss: 0.289963 	Validation Loss: 0.400423
Epoch: 88 	Training Loss: 0.282921 	Validation Loss: 0.421267
Epoch: 89 	Training Loss: 0.284123 	Validation Loss: 0.403326
Epoch: 90 	Training Loss: 0.282097 	Validation Loss: 0.411788
Epoch: 91 	Training Loss: 0.285459 	Validation Loss: 0.424447
Epoch: 92 	Training Loss: 0.287915 	Validation Loss: 0.407220
Epoch: 93 	Training Loss: 0.278051 	Validation Loss: 0.400814
Epoch: 94 	Training Loss: 0.291190 	Validation Loss: 0.424086
Epoch: 95 	Training Loss: 0.271764 	Validation Loss: 0.429150
Epoch: 96 	Training Loss: 0.284526 	Validation Loss: 0.394954
Epoch: 97 	Training Loss: 0.278233 	Validation Loss: 0.421139
Epoch: 98 	Training Loss: 0.276778 	Validation Loss: 0.430977
Epoch: 99 	Training Loss: 0.268164 	Validation Loss: 0.417392
Epoch: 100 	Training Loss: 0.273684 	Validation Loss: 0.417860

<All keys matched successfully>

(IMPLEMENTATION) Test the Model

Try out your model on the test dataset of dog images. Use the code cell below to calculate and print the test loss and accuracy. Ensure that your test accuracy is greater than 60%.

test(loaders_transfer, model_transfer, criterion_transfer, use_cuda)

Test Loss: 0.478605


Test Accuracy: 85% (716/836)

(IMPLEMENTATION) Predict Dog Breed with the Model

Write a function that takes an image path as input and returns the dog breed (Affenpinscher, Afghan hound, etc) that is predicted by your model.

### TODO: Write a function that takes a path to an image as input
### and returns the dog breed that is predicted by the model.

# list of class names by index, i.e. a name can be accessed like class_names[0]
class_names = [item[4:].replace("_", " ") for item in dataset['train'].classes]

def predict_breed_transfer(img_path, model):
    # load the image and return the predicted breed
    img = Image.open(img_path)
    
    tensor = transform(img)
    tensor = tensor.unsqueeze(0)
    
    if use_cuda:
        tensor = tensor.cuda()
        model = model.cuda()
        
    output = model_transfer(tensor)
    dog_class = torch.argmax(output)
    
    plt.axis('off')
    plt.imshow(img)
    plt.show()
    
    return class_names[dog_class.item()]

predict_breed_transfer('WhatsApp Image 2021-06-18 at 21.49.18.jpeg', model_transfer)

'Parson russell terrier'

---

Step 5: Write your Algorithm

Write an algorithm that accepts a file path to an image and first determines whether the image contains a human, dog, or neither. Then,

• if a dog is detected in the image, return the predicted breed.
• if a human is detected in the image, return the resembling dog breed.
• if neither is detected in the image, provide output that indicates an error.

You are welcome to write your own functions for detecting humans and dogs in images, but feel free to use the face_detector and dog_detector functions developed above. You are required to use your CNN from Step 4 to predict dog breed.

Some sample output for our algorithm is provided below, but feel free to design your own user experience!

(IMPLEMENTATION) Write your Algorithm

### TODO: Write your algorithm.
### Feel free to use as many code cells as needed.

def run_app(img_path):
    ## handle cases for a human face, dog, and neither
    if face_detector(img_path):
        print('This human look like a {}'.format(predict_breed_transfer(img_path, model_transfer)))
    elif dog_detector(img_path):
        print('This dog is a {}'.format(predict_breed_transfer(img_path, model_transfer)))
    else:
        print('This picture has no dog or human')

---

Step 6: Test Your Algorithm

In this section, you will take your new algorithm for a spin! What kind of dog does the algorithm think that you look like? If you have a dog, does it predict your dog's breed accurately? If you have a cat, does it mistakenly think that your cat is a dog?

(IMPLEMENTATION) Test Your Algorithm on Sample Images!

Test your algorithm at least six images on your computer. Feel free to use any images you like. Use at least two human and two dog images.

Question 6: Is the output better than you expected :) ? Or worse :( ? Provide at least three possible points of improvement for your algorithm.

Answer: (Three possible points for improvement)

## TODO: Execute your algorithm from Step 6 on
## at least 6 images on your computer.
## Feel free to use as many code cells as needed.
run_app('WhatsApp Image 2021-06-18 at 21.49.18.jpeg')
run_app('WhatsApp Image 2021-06-18 at 18.47.01.jpeg')
run_app('csm_pabllo-vittar_e55e59a27c.jpg')
run_app('edab9ae0a2884159d7e17d5814076fc4b51f761c.jpg')
run_app('6g7ge66rpag51.jpg')
run_app('164742v800_puppy-1.jpg')
## suggested code, below
for file in np.hstack((human_files[:3], dog_files[:3])):
    run_app(file)

This dog is a Parson russell terrier

This dog is a Chihuahua

This human look like a Afghan hound

This dog is a Akita
This picture has no dog or human

This dog is a American staffordshire terrier

This human look like a American water spaniel

This human look like a American water spaniel

This human look like a Silky terrier

This dog is a Tibetan mastiff

This dog is a Tibetan mastiff

This dog is a Tibetan mastiff